← Back to the Labs tiles

New labs 2025–26

7 labs Generated from this page's data. Click a name to open its panel on the Labs page. Records resting on one source are tagged "one source".

This division lists AI labs set up in 2025 and 2026. Cursive, founded by senior DeepMind alumni, builds AI agents that keep learning from real-world use (per the site), and was given public compute by the government's Sovereign AI Unit in April 2026. General Instinct, set up by a DeepMind alumnus, builds vision-language AI models (per the site). The Fundamental AI Research Lab is a government-funded lab, with up to £40 million over six years, for long-term, high-risk research into basic flaws of AI, chosen through a UKRI funding call.

Context

  1. The government announced a new lab for high-risk fundamental AI research, with up to £40 million over six years. 4 March 2026 gov.uk/government/news/government-to-create-new-Located in the UK, up to £40 million will be made available over 6 years.
  2. Cursive was among six start-ups given AI Research Resource compute in Sovereign AI's first round of backing. 16 April 2026 gov.uk/government/news/ai-firms-pioneering-drug-developing AI agents that learn and improve continuously from real-world use
  3. Sovereign AI's first equity investment went to Callosum, an AI infrastructure start-up. 16 April 2026 gov.uk/government/news/ai-firms-pioneering-drug-Sovereign AI's first equity investment will be in the AI infrastructure startup Callosum.
  4. DSIT's AI sector study counted 10,488 UK AI companies in 2025, up 79% on 2024. 29 October 2026 gov.uk/government/publications/artificial-intellA total of 10,488 AI companies were identified, up 79% on 2024.
  5. The AI Opportunities Action Plan recommended a UK Sovereign AI unit able to partner with private companies. 13 January 2025 gov.uk/government/publications/ai-opportunities-Create a new unit, UK Sovereign AI, with the power to partner with the private sector.

Labs Round 8 (31 Oct 2026): introduction and facts as the researchers wrote them; "(per the site)" marks a sentence resting on this site's own description of that lab.

RecordWhat it isPlaceLinksConfidence
CursiveCompaniesLab founded by senior DeepMind alumni building AI agents that keep learning from real-world use; described as a foundation model start-up.
Profile
Founded
2025 (Tech Funding News, trade, single)
Founders
Talfan Evans — co-founder and CEO; Olivier Henaff — co-founder (name from Tech Funding News only)
Chief executive
Talfan Evans — from null
Hq
London (England: London) (Tech Funding News); other_offices: null
What it does
A London lab founded by former DeepMind staff, building AI agents that keep learning from real-world use, using foundation models for predicting behaviour and reasoning over long tasks. It was described as still in stealth in April 2026.
Research focus
continual learning; self-improving agents; behavioural prediction; long-horizon reasoning
Compute
Sovereign AI / AIRR compute, one of six firms — 16 April 2026
Funding total as published
null (investor named by Tech Funding News: SuperSeed; amount undisclosed)
Uk government links
Sovereign AI / AIRR compute — 16 April 2026 (already DEAL-LAB-2035)
Status now
Active London lab in stealth, using Sovereign AI compute since April 2026.
What it does (Labs Round 8)
Not stated (no citable source opened)
Awards and programmes
Sovereign AI / AI Research Resource compute programme — Apr 2026 (already DEAL-LAB-2035; not re-opened this round)
Status now (Labs Round 8)
Not stated — not verified this round

Labs Round 7 and Labs Round 8 deep profile (30–31 Oct 2026), as the researchers wrote it; confidence: one source (nothing verified) ---.

Sources

  1. gov.uk/government/news/ai-firms-pioneering-drug-
  2. sovereignai.gov.uk/post/our-first-investments
  3. telecomtv.com/content/digital-platforms-services
  4. techfundingnews.com/uks-sovereign-ai-fund-names-
Key facts and news (3)
Research focus
Foundation models for behavioural prediction and long-horizon reasoning, so agents improve the more they are deployed
Founders
Olivier Hénaff — co-founder (former DeepMind scientist); Talfan Evans — co-founder one source tech.eu/2026/05/01/over/
UK government links
Access to AIRR supercomputers through Sovereign AI, announced 16 Apr 2026 (see DEAL-LAB-2035)

Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked.

gov.uk/government/news/ai-firms-pioneering-drug- · tech.eu/2026/05/01/over/ · sovereignai.gov.uk/post/our-first-investments · telecomtv.com/content/digital-platforms-servicesverified
Fundamental AI Research Lab (DSIT and UKRI)Schemes, programmes and AI linesA government-funded UK lab for long-term, high-risk research into the basic flaws of AI, chosen through a UKRI funding call. call closed; award pendingKey facts Up to £40m over six years, confirmed 4 Mar 2026. In-kind access to AI Research Resource compute "worth tens of millions of pounds". Applications through a UKRI funding call, peer reviewed; panel chaired by Raia Hadsell (VP of Research, Google DeepMind; DSIT AI ambassador). Aim: rethink how AI is built rather than scale existing systems. Host institution and director not named on pages opened.
Profile
Owner or parent
DSIT and UKRI (EPSRC)
Founded or opened
funding call announced 4 Mar 2026
Lab director or lead
Not stated in the call. Raia Hadsell — VP of Research at Google DeepMind and UK government AI Ambassador — chairs the peer-review panel (Mar 2026)
Other leaders
Dr Kedar Pandya — Executive Director, Strategy Directorate, EPSRC; Kanishka Narayan MP — AI Minister
What it does
A government plan, announced in March 2026, to fund a lab doing high-risk, blue-sky AI research. The lab is to tackle flaws in current models, such as hallucinations, short memory and unpredictability, and to try new architectures instead of only scaling up existing ones. On 23 June 2026 UKRI-EPSRC funded two labs from a £60m six-year scheme: SOFAIR at UCL and BOLD at Oxford. No page I opened says these came from the March call.
Research groups or focus
basic flaws in current models (hallucinations, short memory, unpredictability); more accurate, transparent and trustworthy AI; new AI architectures
Funding or budget as published
up to £40m over six years, plus in-kind AI Research Resource compute "worth tens of millions of pounds" — 4 Mar 2026 — https://www.gov.uk/government/news/government-to-create-new-lab-to-keep-uk-in-the-fast-lane-on-ai-breakthroughs (already DEAL-LAB-6002). Related: up to £60m over six years for SOFAIR and BOLD (SOFAIR's initial share £8m) — 23 Jun 2026 (new deal DEAL-LAB-15001)
Partners named
SOFAIR Lab — led by Professor David Barber (UCL), with Cambridge, Oxford and Edinburgh — Jun 2026; BOLD Lab — led by Jakob Foerster (Oxford), with UCL and Imperial — Jun 2026; NVIDIA — compute for both labs within its $250m UK commitment — Oct 2026
Uk government links
DSIT and UKRI; AIRR compute; launched under UKRI's AI Strategy
Status now
The call closed into two funded labs announced on 23 Jun 2026 (SOFAIR at UCL, BOLD at Oxford), although no opened page explicitly ties them to this call.

Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: verified (call: GOV.UK and UKTN; SOFAIR/BOLD funding: UCL and UKTN); the link between the call and the two labs is unverified.

Sources

  1. gov.uk/government/news/government-to-create-new-
  2. uktech.news/ai/uk-backs-new-ai-research-labs-in-
  3. ucl.ac.uk/news/2026/jun/national-research-lab-ba
  4. nvidianews.nvidia.com/news/nvidia-commits-250-mi
Key facts and news (5)
Status now
Call closed 31 Mar 2026; interviews 16–17 Apr 2026; stated lab start 1 May 2026; stage-gate outcomes aimed for early Feb 2027; one award only; no winning consortium named on the UKRI page as read 31 Oct 2026 ukri.org/opportunity/fundamental-ai-research-lab
Funding as published
Up to £40m in stages over six years: initial 18-month research grant up to £7.5m full economic cost (EPSRC pays 80%); training grant up to £1.9m for at least 10 doctoral students; up to 2 million GPU hours a year of AIRR compute ukri.org/opportunity/fundamental-ai-research-lab
Research focus
Novel fundamental AI not pursued by commercial labs; trade report names hallucinations, unreliable memory and unpredictable reasoning
UK government links
Funded by UKRI (EPSRC) with DSIT; AIRR compute; peer review panel chaired by Raia Hadsell (Google DeepMind VP of Research; DSIT AI Ambassador) one source capacityglobal.com/news/uk-launches-40m-moonshot

News

Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked.

gov.uk/government/news/government-to-create-new- · uktech.news/ai/uk-government-to-launch-new-40m-a · ukri.org/opportunity/fundamental-ai-research-lab · capacityglobal.com/news/uk-launches-40m-moonshotverified
General InstinctCompaniesStart-up set up by a DeepMind alumnus to build vision-language AI models.
Profile
What it does
Not stated (no citable source opened)
Status now
Not stated — not verified (a directory lists it as inactive; see Gaps)

Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: one source (nothing verified; record should stay thin until a primary source is found); sources are listed in the record panel.

Key facts and news (2)
Research focus
Vision-language models one source tech.eu/2026/05/01/over/
Founders
Guanming Wang — founder (former DeepMind intern, per tech.eu) one source tech.eu/2026/05/01/over/

Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked.

tech.eu/2026/05/01/over/one source
Ineffable IntelligenceCompaniesResearch company set up by David Silver to build an AI system that learns skills and knowledge from its own experience rather than from human data.
Profile
Legal name
Ineffable Intelligence Ltd (website terms; registered office stated as Altrincham, Cheshire)
Founded
null (David Silver's note on setting up the company dated 15 Jan 2026)
Founders
David Silver — founder and CEO
Chief executive
David Silver — from null
Chief scientist or research lead
David Silver (professor at University College London; formerly led the reinforcement learning team at Google DeepMind)
Hq
London (England: London); other_offices: null
What it does
AI research lab building a "superlearner" — an AI system that learns from its own experience through reinforcement learning rather than from human data — with the stated aim of "making first contact with superintelligence".
Research focus
large-scale reinforcement learning; systems that learn continuously from experience
Models and products
null (none announced)
Customers or partners named
NVIDIA — engineering collaboration on RL infrastructure — 13 May 2026 — https://www.ineffable.ai/blog/nvidia-ineffable-intelligence-team-up-to-build-the-future-of-reinforcement-learning-infrastructure; Google Cloud — preferred cloud partner — 16 Jun 2026 — https://www.uktech.news/ai/ineffable-intelligence-and-google-cloud-partner-on-new-frontier-ai-lab-20260616
Compute
Co-designing RL training infrastructure with NVIDIA, starting on Grace Blackwell and among the first to explore Vera Rubin (13 May 2026); Google Cloud chosen as preferred cloud partner for RL infrastructure (16 Jun 2026)
Funding total as published
1,100,000,000 USD (£814m) seed — 28 Apr 2026 — (trade) UKTN
Valuation reported
null ("not disclosed" per UKTN)
Rounds found
2026-04-28 — seed — 1,100,000,000 USD (£814m) — lead not named by UKTN (site holds Sequoia and Lightspeed as co-leads, DEAL-LAB-025) — Sovereign AI Unit (co-investing with British Business Bank), British Business Bank ($20m separately), Sequoia, Lightspeed, NVIDIA, Index, Google, EQT, Evantic, Flying Fish, DST Global, BOND
Uk government links
Sovereign AI Unit co-invested with the British Business Bank in the seed round — 28 Apr 2026
Status now
Emerged from stealth in spring 2026 with a $1.1bn seed round and compute partnerships with NVIDIA and Google Cloud; no model released.

Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: single_source (funding trade-only; company site does not state it).

Sources

  1. ineffable.ai/
  2. ineffable.ai/blog/nvidia-ineffable-intelligence-
  3. ineffable.ai/terms-and-conditions
Why it is here: came out of stealth on 27 Apr 2026 with a $1.1bn seed round
Key facts and news (4)
Valuation (reported)
$5.1bn at seed (Apr 2026), as stated by the company one source tech.eu/2026/04/27/ineffable-intelligence-launch
Research focus
Reinforcement learning: a "superlearner" that keeps discovering knowledge and skills without relying on human data tech.eu/2026/04/27/ineffable-intelligence-launch
Research leads
Wojciech Marian Czarnecki — chief scientific officer (joined by May 2026) one source tech.eu/2026/05/01/over/
Funding as published
$1.1bn seed (Apr 2026) co-led by Sequoia and Lightspeed; participants NVIDIA, DST Global, Index, Google, Flying Fish Ventures, EQT Ventures, Evantic Capital, UK Wellcome Trust, BOND Capital, British Business Bank, Sovereign AI Fund tech.eu/2026/04/27/ineffable-intelligence-launch

Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked.

(city not published)Companies House · tech.eu/2026/04/27/ineffable-intelligence-launch · uktech.news/ai/ineffable-intelligence-secures-81 · ineffable.ai/ · find-and-update.company-information.service.gov.verified
InherentCompaniesLondon research lab building Faraday, an AI system meant to let scientists and self-improving AI work together on research.
Profile
Legal name
null (site footer reads "© 2026 Inherent Laboratories"; the site says it is incorporated as a Public Benefit Corporation; Companies House not checked)
Founders
Tantum Collins — Co-Founder & CEO; Edward Hughes — Co-Founder & Chief Scientist; Louis Kirsch — Co-Founder; Kaloyan Aleksiev — Co-Founder & CTO
Chief executive
Tantum Collins — from null
Chief scientist or research lead
Edward Hughes — Co-Founder & Chief Scientist
Other leaders
Matt Clifford — adviser (tech.eu)
Hq
London (England: London); other_offices: null
What it does
A London AI research lab building "general-purpose inventive AI" to speed up scientific research. Its system, Faraday, is meant to let people and self-improving AI work together on hard scientific problems. It says it applies recursive self-improvement to its models and to how it runs research.
Research focus
recursive self-improvement; AI Scientist agents; open-ended research; human–machine teaming; autonomous labs
Models and products
Faraday — AI system for scientific discovery — announced May 2026; reported in August 2026 to run on Qwen 3.6 with 27 billion parameters (trade, single)
Papers and awards
"Training AI Scientists to Replicate Research" — 14 August 2026 (inherentlabs.ai/research)
Funding total as published
50,000,000 USD — as of 29 May 2026 — https://tech.eu/2026/05/29/london-based-ai-lab-inherent-emerges-from-stealth-with-50m-raise/
Rounds found
2026-05-29 — seed — 50,000,000 USD — Index Ventures and Radical Ventures (co-leads) — none named (already DEAL-LAB-027 / DEAL-LAB-2009)
Uk government links
null (tech.eu describes adviser Matt Clifford as a former government AI adviser)
Status now
Active London research lab that left stealth in May 2026 and published its first research post in August 2026.

Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: verified (founders, roles, round: investor page plus trade); Faraday model details single_source.

Sources

  1. radical.vc/portfolio/inherent/
  2. tech.eu/2026/05/29/london-based-ai-lab-inherent-
  3. inherentlabs.ai/
  4. inherentlabs.ai/research/
  5. newsbytesapp.com/news/science/deepmind-alumni-s-
Models and products (1)
  1. Faraday (Inherent AI scientist agent) 2026-08-14 research releaseInherent's AI "scientist" agent, trained with reinforcement learning to reproduce the findings of published research papers.
Why it is here: London AI lab that came out of stealth on 29 May 2026
Key facts and news (2)
Founders
Tantum Collins, Edward Hughes, Louis Kirsch (all formerly DeepMind; Collins also White House AI policy); Kaloyan Aleksiev (formerly Reka AI and Microsoft) tech.eu/2026/05/29/london-based-ai-lab-inherent-
Status now
Active; set up as a public benefit corporation (May 2026) one source uktech.news/ai/london-based-ai-lab-inherent-emer

Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked.

LondonCompanies House · tech.eu/2026/05/29/london-based-ai-lab-inherent- · uktech.news/ai/london-based-ai-lab-inherent-emer · inherentlabs.ai/ · find-and-update.company-information.service.gov.verified
Metis ReasoningCompaniesNew company developing an AI reasoning system meant to be easier to check and to need far less computing power than today's large models.
Profile
Legal name
null (site footer "© 2026 Metis Reasoning")
Founded
null (public launch 20 October 2026)
Founders
Thore Graepel — co-founder (AlphaGo co-creator; left Google DeepMind summer 2026); Piers Ingram — co-founder; Michael Warren — co-founder (robotics, ex-Boeing); David Berman — co-founder (Head of AI, Cambridge Consultants)
Hq
London (England: London); other_offices: Singapore; Los Angeles (as listed in the white paper, per The Next Web)
Headcount
12 — October 2026 — https://tech.eu/2026/10/20/former-google-deepmind-top-researcher-launches-london-compute-efficient-ai-company/ (trade; The Next Web repeats tech.eu)
What it does
A London lab working on AI reasoning that it says needs much less compute than frontier large language models. It wants systems that keep an auditable record of what they know and how sure they are, for high-stakes work in robotics, science and engineering. It has published a white paper but no model or test results.
Research focus
"generalizable machine reasoning"; embodied AI, science and engineering; approach drawn from AlphaGo/AlphaZero self-play and Jaakko Hintikka's inquirer-versus-nature game (white paper, per The Next Web)
Models and products
The Age of Reasoning — white paper — October 2026; no model released
Papers and awards
"The Age of Reasoning" white paper — October 2026
Compute
says its design is "much much more compute-efficient" than frontier models — October 2026
Funding total as published
null (reported to be raising tens of millions of dollars; the company would not confirm the amount)
Uk government links
David Berman described as a UK government and NATO AI adviser (tech.eu) — October 2026
Status now
Launched publicly in London on 20 October 2026; no product or results released.

Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: single_source (founders and headcount from trade only; the company site names no people).

Sources

  1. tech.eu/2026/10/20/former-google-deepmind-top-re
  2. thenextweb.com/news/metis-reasoning-thore-graepe
  3. metisreasoning.ai/
Why it is here: London AI company launched publicly on 20 Oct 2026
Selected papers (1)
  1. The Age of Reasoning: Intelligence from first principles (Graepel, Berman, Warren, Ingram) Metis Reasoning white paper · Oct 2026
Key facts and news (7)
Founders
Founding team: Thore Graepel (former Google DeepMind research lead, AlphaGo co-creator); Piers Ingram (technology and biotech entrepreneur); Michael Warren (robotics and autonomous systems, formerly Boeing); David Berman (head of AI at Cambridge Consultants, per tech.eu) tech.eu/2026/10/20/former-google-deepmind-top-re
Headcount
12 (Oct 2026) one source
Offices
London; Singapore; Los Angeles one source metisreasoning.ai/whitepaper/the-age-of-reasonin
What it does
A new approach to generalisable machine reasoning, aimed at embodied AI, science and engineering, where correctness is critical metisreasoning.ai/
Research focus
Reasoning framed as a "game of inquiry" (modelled on AlphaGo); learned reasoning strategies advance an explicit, persistent epistemic state, keeping knowledge separate from reasoning for auditability metisreasoning.ai/whitepaper/the-age-of-reasonin
Status now
Launched 20 Oct 2026; reported to be raising "tens of millions of dollars" (amount not confirmed by the company); timing of first public model not set one source

News

Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked.

LondonCompanies House · tech.eu/2026/10/20/former-google-deepmind-top-re · metisreasoning.ai/ · find-and-update.company-information.service.gov. · find-and-update.company-information.service.gov.verified
Recursive SuperintelligenceCompaniesResearch company building AI systems that improve themselves, starting by automating AI research and the scientific method.
Profile
Founded
null (emerged from stealth 13 May 2026)
Founders
Richard Socher — leads the company "alongside seven co-founders" (former chief scientist at Salesforce, founder of You.com); Yuandong Tian — co-founder (former research scientist director, Meta FAIR); Tim Rocktäschel — co-founder (Professor of AI, UCL; former principal scientist, Google DeepMind); Alexey Dosovitskiy — co-founder; Josh Tobin — co-founder (formerly OpenAI); Caiming Xiong — co-founder; Tim Shi — co-founder; Jeff Clune — co-founder
Chief executive
Richard Socher — leads the company (title not stated in source)
Other leaders
Peter Norvig — adviser
Hq
London (England: London) per BusinessCloud; other_offices: San Francisco
Headcount
more than 25 — May 2026 — BusinessCloud and TNW (both trade)
What it does
AI research company building systems that run their own experiments to improve themselves safely — "recursive self-improvement" through open-ended automated scientific discovery.
Research focus
recursive self-improvement; automated scientific discovery
Compute
null (NVIDIA and AMD are investors)
Funding total as published
650,000,000 USD (£480m) — May 2026 — trade
Valuation reported
4,650,000,000 USD — 14 May 2026 — (trade) TNW; about £3.5bn — 14 May 2026 — (trade) BusinessCloud (two trade sources; no primary opened)
Rounds found
2026-05-13 — first round (TNW: emerged from stealth; other listings call it Series A — not verified) — 650,000,000 USD / £480m — GV, Greycroft — NVIDIA, AMD (Ventures)
Status now
Emerged from stealth in May 2026 with $650m at a reported $4.65bn valuation; offices in London and San Francisco.

Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: single_source (two trade sources, no primary); sources are listed in the record panel.

Models and products (1)
  1. Recursive Flint 2026 (date not given on page) releasedRecursive Superintelligence's first public model: a 40-billion-parameter coding model built to run fast on local machines.
Why it is here: incorporated in London; came out of stealth 13 May 2026
Key facts and news (5)
Valuation (reported)
$4.65bn (May 2026) one source tech.eu/2026/05/13/recursive-superintelligence-e
Headcount
Fewer than 30 (May 2026) one source tech.eu/2026/05/13/recursive-superintelligence-e
Offices
London; San Francisco one source
Research focus
AI systems that improve themselves by analysing their own performance, without human intervention one source
Funding as published
Over $650m (May 2026) led by GV and Greycroft, with Nvidia and AMD participating

Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked.

King's Cross, London (N1 9FN)Companies House · tech.eu/2026/05/13/recursive-superintelligence-e · find-and-update.company-information.service.gov. · find-and-update.company-information.service.gov. · recursive.com/one source

Tile photo: King's Cross Central development Coal Drops Yard and Gasholders, London England 03, Acabashi, CC BY-SA 4.0 (Wikimedia Commons).